Feature-Specific Profiling

Abstract

While high-level languages come with signi cant readability and maintainability bene ts, their performance remains di cult to predict. For example, programmers may unknowingly use language features inappropriately, which cause their programs to run slower than expected. To address this issue, we introduce feature-speci c pro ling, a technique that reports performance costs in terms of linguistic constructs. Festure-speci c pro lers help programmers nd expensive uses of speci c features of their language. We describe the architecture of a pro ler that implements our approach, explain prototypes of the pro ler for two languages with di erent characteristics and implementation strategies, and provide empirical evidence for the approach's general usefulness as a performance debugging tool.